DeepRL2.1, Introduction: Mini-Batches in On- and Off-Policy Deep Reinforcement Learning

views comments

One of the challenges in Deep Reinforcement Learning is to decorrelate the data. How this is possible with replay buffers is explained here.